Deciphering the microbial complexity of Chinese traditional sourdough through integrated learning

Area of Science:

• Microbiology
• Food Science
• Bioinformatics

Background:

• Sourdough fermentation is driven by complex microbial consortia (fungi and bacteria).
• Interspecies interactions and environmental factors significantly influence sourdough microbiota composition.
• Understanding these relationships is crucial for controlling fermentation and product quality.

Purpose of the Study:

• To analyze the microbiota and physicochemical properties of diverse sourdough samples from China.
• To correlate microbial composition with physicochemical characteristics using integrated learning approaches.
• To identify key microbial players and environmental factors shaping sourdough fermentation.

Main Methods:

• Analysis of 115 sourdough samples (26 dry, 89 wet) from ten Chinese provinces.
• Microbiota profiling and physicochemical characterization.
• Application of ensemble learning and abundance ranking for data integration.

Main Results:

• Distinct microbial communities were observed in dry (Pediococcus pentosaceus, Levilactobacillus brevis) and wet (Fructilactobacillus sanfranciscensis) samples.
• Saccharomyces cerevisiae was a dominant and core yeast species across samples.
• Rare bacterial taxa showed significant correlations with physicochemical indices like pH and acidity.
• Microbial structure strongly correlated with moisture content, pH, and total acidity.

Conclusions:

• Microbial community structure is closely linked to physicochemical conditions in sourdough.
• Machine learning can uncover critical microbial-environment relationships in fermentation.
• Findings support the rational design of stable starter cultures for optimized sourdough production.